[ZeroMQ] The new solution for building up distributed system

    I have used TIBCO Rendezvous (RV Message), ActiveMQ (JMS), and RabbitMQ(AMQP) before. They all have message broker, which means all the messages will be send to broker (centralized control) first, and then the client or node will receives them later. But, ZeroMQ has totally different story. I personally believe that it could be a very good solution for building up distributed system.

http://www.zeromq.org/

    And also, this article gives the short introduction and summarizes the important items including "communication transport", "End Point Implementation", "The Socket Object", and so on about ZeroMQ. This content is very uesful for beginner to get to know and read first.
http://www.coastrd.com/zeromq-messaging

[Python] Make pyc file for your Python source code

It could be a situation when you have to give your Python program to customers, but you don't want to give them Python source code. Here is a solution. Give them Python byte code!

For example, on command line:

• Compile one file

         > python -c "import compileall; compileall.compile_file('YourPythonFile.py')"

• Compile one folder

         > python -c "import compileall; compileall.compile_dir('YourFolder/', force=True)"

Or, you can put the compiling action in source code

      import compileall
      compileall.compile_dir('YourFolder/', force=True)

[Quantum] An example of OpenStack Quantum's table schema and data

mysql> SELECT * FROM ovs_quantum.networks n LIMIT 0,1000;
+--------------------------------------+-----------+---------+-----------+
| uuid                                 | tenant_id | name    | op_status |
+--------------------------------------+-----------+---------+-----------+
| 5b2c8537-26df-4fdc-9e38-3f3f09797d3f | default   | private | UP        |
+--------------------------------------+-----------+---------+-----------+
1 row in set (0.00 sec)
mysql> SELECT * FROM ovs_quantum.ports p LIMIT 0,1000;
+--------------------------------------+--------------------------------------+--------------------------------------+--------+-----------+
| uuid                                 | network_id                           | interface_id                         | state  | op_status |
+--------------------------------------+--------------------------------------+--------------------------------------+--------+-----------+
| e1b7cec5-1d7c-4bcd-b123-4a97f37ef498 | 5b2c8537-26df-4fdc-9e38-3f3f09797d3f | 0a76a701-ea3b-40d8-a56d-fc26a8db69c2 | ACTIVE | DOWN      |
| fdc3836d-c884-4aeb-b5b2-e354745a2de4 | 5b2c8537-26df-4fdc-9e38-3f3f09797d3f | gw-5b2c8537-26                       | ACTIVE | DOWN      |
| d612ccc3-81cd-489c-8cf6-3e37966cebcb | 5b2c8537-26df-4fdc-9e38-3f3f09797d3f | b91861ff-f90b-435d-b9c2-65eb76f0300d | ACTIVE | DOWN      |
+--------------------------------------+--------------------------------------+--------------------------------------+--------+-----------+
3 rows in set (0.00 sec)
mysql> SELECT * FROM ovs_quantum.vlan_bindings v LIMIT 0,1000;
+---------+--------------------------------------+
| vlan_id | network_id                           |
+---------+--------------------------------------+
|       2 | 5b2c8537-26df-4fdc-9e38-3f3f09797d3f |
+---------+--------------------------------------+
1 row in set (0.00 sec)

[sFlow] sFlow Agent and sFlow Collector

sFlow is a technology for monitoring network, wireless and host devices. Based on the following topology, there is a test about using sFlow Agent and sFlow Collector to observe the sFlow data from Open vSwitch.

• sFlow agent is from Open vSwitch

          How to setup sFlow on Open vSwitch:
          > sudo ovs-vsctl -- --id=@s create sFlow agent=eth0 target=\"colletor ip:6343\" header=128 sampling=64 polling=10 -- set Bridge br0 sflow=@s 

• sFlow collector is sFlowTrend

 Now, we can see the results from sFlowTrend:

 When trying to ping PC1 and PC2, the result is below:

[Comparsion] Topology graph in NOX GUI and OpenFlow GUI

Using GNS3 for constructing an emulation network environment, I give a simple topology as follows to try running NOX GUI and OpenFlow GUI.
P.S: for more information in GNS3, please refer to this:
http://www.gns3.net/gns3-virtualbox-edition/

In NOX GUI, the topology graph is created like this:

In OpenFlow GUI, the topology graph is created like this:

[Trema] Try Trema App: flow_dumper

1. Choose "routing_switch" as my testing app. (any other simple switch app is all available...)
> ./trema run -c ../apps/routing_switch/routing_switch.conf

2. Send packets between these 2 hosts
> ./trema send_packets -s host2 -d host1
> ./trema send_packets -s host1 -d host2

3. Try to run Trema App: "flow_dumper" as follows
> TREMA_HOME=. ../apps/flow_dumper/flow_dumper
[0x000000000000e0] priority = 65535, match = [wildcards = 0, in_port = 1, dl_src = 00:00:00:01:00:02, dl_dst = 00:00:00:01:00:01, dl_vlan = 65535, dl_vlan_pcp = 0, dl_type = 0x800, nw_tos = 0, nw_proto = 17, nw_src = 192.168.0.2/32, nw_dst = 192.168.0.1/32, tp_src = 1, tp_dst = 1], actions = [output: port=2 max_len=65535]
[0x000000000000e0] priority = 65535, match = [wildcards = 0, in_port = 2, dl_src = 00:00:00:01:00:01, dl_dst = 00:00:00:01:00:02, dl_vlan = 65535, dl_vlan_pcp = 0, dl_type = 0x800, nw_tos = 0, nw_proto = 17, nw_src = 192.168.0.1/32, nw_dst = 192.168.0.2/32, tp_src = 1, tp_dst = 1], actions = [output: port=1 max_len=65535]

4. So, we can see all the flows here.

[Tutorial] How to setup QoS on Open vSwitch

There are two ways to do that:

1. Interface Rate Limiting ( on Interface )

• For instance:

         > sudo ovs-vsctl set Interface eth1 ingress_policing_rate=10000
         > sudo ovs-vsctl set Interface eth1 ingress_policing_burst=1000

 2. Port  QoS Policy ( on Port )

• For instance:

         > sudo ovs-vsctl set port eth1 qos=@newqos \
             -- --id=@newqos create qos type=linux-htb \
             other-config:max-rate=200000000 queues=0=@q0,1=@q1 \
             -- --id=@q0 create queue \
             other-config:min-rate=100000000 \
             other-config:max-rate=100000000 \
              -- --id=@q1 create queue \
             other-config:min-rate=50000000 \
             other-config:max-rate=50000000

• Qos can have more than 1 queue

[Tutorial] Trema Tutorial

http://www.fp7-ofelia.eu/assets/Uploads/201203xx-TremaTutorial.pdf

[Open vSwitch] How to get port statistics from interface in OVS

1. Show the bridge info
> sudo ovs-vsctl show
result:
2909bfce-536e-4184-a5bb-507f0553abee
    Bridge "br0"
        Controller "tcp:10.6.186.244"
        Port "br0"
            Interface "br0"
                type: internal
        Port "eth3"
            Interface "eth3"
        Port "eth2"
            Interface "eth2"
        Port "eth1"

2. Get port statistics from interface
> sudo ovs-vsctl get Interface br0 statistics
result:

{collisions=0,

rx_bytes=0,

rx_crc_err=0,

rx_dropped=0,

rx_errors=0,

rx_frame_err=0,

rx_over_err=0,

rx_packets=0,

tx_bytes=0,

tx_dropped=0,

tx_errors=0,

tx_packets=0}

[Python] How to use Decorators for Functions and Methods

When list_ports() is executed, Class ApiCall will be created for becoming a decorator to examine the arguments as follows:

import sys class ApiCall(object): """A Decorator to add support for format and tenant overriding""" def __init__(self, function): self.function = function def __get__(self, instance, owner): def with_params(*args, **kwargs): """ Temporarily sets the format and tenant for this request """ (format, tenant) = (instance.format, instance.tenant) if 'format' in kwargs: instance.format = kwargs['format'] if 'tenant' in kwargs: instance.tenant = kwargs['tenant'] ret = self.function(instance, *args) (instance.format, instance.tenant) = (format, tenant) return ret return with_params class Client(object): def __init__(self, tenant=None, format="xml"): self.tenant = tenant self.format = format @ApiCall def list_ports(self, network): """ Fetches a list of ports on a given network """ return network def main(): client = Client(tenant="AAA",format="xml") client.list_ports('my network') sys.exit(0) if __name__ == "__main__": main()